Is it correct that the "flat, even anechoic response" translates to the sloping downward response seen in the predicted in-room measurements?
Would you be able to sketch or show the target curve with the elevated mid-bass that you mention? Or are the following near enough...
View attachment 278980
Both curves have a slope of 1dB/octave above 500Hz (I've assumed that an approximately 1dB/octave slope, across the whole range, occurs when you convert from anechoic to predicted in-room). (The target curves above would be intended to be applied to the in-room response rather than the anechoic one).
Below 500Hz, since you state a 1 - 2 dB slope I've added an additional slope of 1.5dB/octave to the 1dB/octave from conversion to PIR. This gives a 2.5dB/octave slope to the roll-off to (hopefully) show how it would appear on the predicted in-room graph. (Is this what you intended)?
The red curve is supposed to show your suggested target for a bass reflex, passive speaker, intended to be used full range with a bass roll-off at 20Hz.
The green curve shows a similar speaker's response, but this time it is intended to be used with a sub crossed at 80Hz.
(Is this approximately the target that you would use for such a speaker)?
What additional considerations might make you modify the target (particularly in the design intended for use with a sub)?